The Executive Brief: Intel SGX Hack Was a Distraction.
Recently Intel’s SGX security feature was found to be susceptible to Foreshadow and Foreshadow-NG attacks. Attackers may use speculative execution, and an unprotected copy of the SGX protected memory to steal sensitive data on unpatched systems. The attack broke the threat model SGX proposed: malicious software running on the CPU may not gain access to SGX protected space.
Does it matter?
For most of us, if there’s malicious software that’s running on our computers, we’ve got bigger problems.
But for some, it matters a great deal. Fortanix, for example, is building an “HSM as a Service” on top of Intel’s SGX feature. When making those types of security claims, yes, it matters a great deal.
But there’s more to security than just correctness. The 2019 Thales Data Threat Report surveyed hundreds of companies and found that,
the number one perceived barrier to implementing data security was complexity.
But just because modern data ecosystems are complex (considering IaaS, PaaS, and SaaS in running on-prem, in-cloud, and hybrid deployments) doesn’t mean the data security layer needs to be complicated.
This disconnect between numerous security practitioner’s values, and, the market’s needs is why security is often a painfully expensive investment. It’s the cost paid for misaligned values, and the SGX hack just a small line-item in that amortized cost.
What can we do about it?
(1) Focus on Simplicity: There are companies out there whose primary value proposition is high-level data security solutions. For example, Peacemakr offers the first Secure Data Platform. They provide an entire platform to build secure data solutions that span vast data ecosystems. Solving all the hard problems correctly with the simplest interface.
(2) Eliminate Information Asymmetries: SGX was not a standard, and the feature was only kept secure via an asymmetry in information. A hacker’s job is to find and exploit information asymmetries. Protect your organization by avoiding building on top of a black box of which you have no visibility.
Vendors that provide open-source level transparency may reduce the possibility of secret information asymmetries being discovered and exploited.
(3) Give Your Org Visibility: Secure data needs to flow from one authorized system to another, seamlessly, regardless of CPU architecture. However, it should never flow between unauthorized systems. Akita Software provides visibility into your dataflows so that you can find leaks faster.
(4) Collaborate Towards Open Standards:
Secure data solutions should foster collaboration between you, your partners, and your customers — not hinder it. Standards are technical and have a better chance of succeeding if they come from a 3rd party. Often, open-source can be leverage to lubricate any tensions. Once the adoption of a secure data standard reaches critical mass, the cost of secure collaboration disappears.